How to wire 192 cells? Multiple BMS's? How many cells or strings in parallel is too many in a 48v system?

[aentrop]

I have 48 Valence U27-12XP LiFeMgPO4 batteries. Each battery has 4 cells inside and NO INTERNAL BMS. I have pulled out the balance wires from inside and connected to a JK 16s 200a BMS (4 packs in series, 16s total cells configuration) and this works great for one 16s 48v string. but now I want to add in 11 more strings.

Currently just running a growatt 3k inverter, will expand eventually but for now the biggest load I run is my EV charger (12A @ 120vac) + a few other devices, rarely ever go over 2500watt load.) Also have a Victron Shunt and Cerbo.

I am considering doing a 4s12p setup for the batteries (if you count the cells it would be 16s12p). Would I need to do 12x 16s BMS's all in parallel to run this pack? How else can I monitor all the cells?

In theory I could parallel every single row of cells and run it all through one BMS but the balance wires are small and that doesn't seem safe to have that many cells paralleled.

Whats a good starting point?

 

[time2roll]

I would be looking at 3x 16s4p to find some place in the middle. No real science.

Might be worth getting a cell capacity tester to match cells together by capacity.

 

[aentrop]

I would be looking at 3x 16s4p to find some place in the middle. No real science.

Might be worth getting a cell capacity tester to match cells together by capacity.

I cant break apart the individual cells, each "battery/pack" has 4 cells inside. I do have all of the packs capacity tested and matching . They are all right about 95-105Ah each.

So If I split it into 3 banks (16s4p each), would I use one BMS per bank? Or do I need a separate BMS for each 16s string?

Is it ok to just combine all the 16+1 balancing wires in parallel? They are small 22awg wires that come out of each battery pack for the balance leads.


Here are photos of the batteries I am using showing the balance wires exposed and ready for connection:
https://diysolarforum.com/threads/valence-xp-super-thread.5683/post-576160

 

[Shimmy]

I don't think I would want to do more than two strings in parallel per BMS, and go for 8 BMSs. I think you are asking for problems with 192 balance leads trying to do the job.

 

[time2roll]

Well not going to parallel cells with the balance wires. Yes looking at 11 more BMS unless the cells can be exposed to connect in parallel with a bus or cable to carry real current.

 

[sunshine_eggo]

I have 48 Valence U27-12XP LiFeMgPO4 batteries. Each battery has 4 cells inside and NO INTERNAL BMS. I have pulled out the balance wires from inside and connected to a JK 16s 200a BMS (4 packs in series, 16s total cells configuration) and this works great for one 16s 48v string. but now I want to add in 11 more strings.

Currently just running a growatt 3k inverter, will expand eventually but for now the biggest load I run is my EV charger (12A @ 120vac) + a few other devices, rarely ever go over 2500watt load.) Also have a Victron Shunt and Cerbo.

I am considering doing a 4s12p setup for the batteries (if you count the cells it would be 16s12p). Would I need to do 12x 16s BMS's all in parallel to run this pack? How else can I monitor all the cells?

In theory I could parallel every single row of cells and run it all through one BMS but the balance wires are small and that doesn't seem safe to have that many cells paralleled.

Whats a good starting point?

Either 12 BMS as you describe or:

Remove Cells
Test cells
Match cells
Build XP16S batteries with 13-X BMS

 

[aentrop]

I don't think I would want to do more than two strings in parallel per BMS, and go for 8 BMSs. I think you are asking for problems with 192 balance leads trying to do the job.

Do you mean 6 BMS's? I have 12p, so split them into 6x 2p's and so I will end up with 6 bms's then paralleled together right?

And when I connect 2 strings in 2p going to one BMS, then I will need to combine in parallel the balance leads for each cell between the two strings right?

 

[aentrop]

Well not going to parallel cells with the balance wires. Yes looking at 11 more BMS unless the cells can be exposed to connect in parallel with a bus or cable to carry real current.

I have terminals on the top for the charge and discharge loads, connecting the balance wires in parallel would only be small deviations between cells. Or are you saying this is still dangerous? I posted a drawing above I hope that helps

 

[sunshine_eggo]

Either 12 BMS as you describe or:

Remove Cells
Test cells
Match cells
Build XP16S batteries with 13-X BMS

OR

Remove cells
Flip them all the same direction in parallel.
Each valence becomes a single 380-420Ah "cell."
Then you would have (4P)16S with a BMS.
Put all three of them together in parallel, and that accounts for 192 cells.

 

[time2roll]

Might work and would be an interesting project. Yes any significant imbalance could overload the small balance leads. Best to put a parallel connection between at the interconnects. That wire is small enough to be considered self fusing. Test in a safe place but I am not really recommending it. Interested to hear the challenges and results.

 

[Shimmy]

Do you mean 6 BMS's? I have 12p, so split them into 6x 2p's and so I will end up with 6 bms's then paralleled together right?

And when I connect 2 strings in 2p going to one BMS, then I will need to combine in parallel the balance leads for each cell between the two strings right?

That is what I was thinking. Not ideal, but saves you so money. 12x BMSs is best; you can parallel 15-16 typically. Fault current gets high with so many strings in parallel; be sure to have a properly rated fuse.

 

[aentrop]

Another decision I have to make.... What size amps BMS's should I go with? I dont see any reason to buy 100+amp bms's. I pulled my history and my max amps draw over the last 6 months has been about 58-60amps, and thats on the current 4s3p bank. So each string is only seeing a max of 20 amps each.

So with that said I am thinking of going with 60 amp JK BMS, BD6A20S8P for $60ea.

Are there any scenarios I might me overlooking or reasons why I need a bigger bms? If I end up with 10 BMS's thats already $600 in just BMS's...

 

[PreppenWolf]

I have 128 cells creating (8) 16s 48V packs. Each with a 100A BMS.

I personally would never mix S/P with individual cells. There is just too much risk of not monitoring each cell voltage at an individual level. If you dump a cell in a 16sXp config, the BMS will not protect the assembly, it will keep chugging along if a cell fully dies.

 

[Shimmy]

So with that said I am thinking of going with 60 amp JK BMS, BD6A20S8P for $60ea.

60A should be plenty; just make sure the BMS can parallel with the number you plan on using.

 

[aentrop]

I have 128 cells creating (8) 16s 48V packs. Each with a 100A BMS.

I personally would never mix S/P with individual cells. There is just too much risk of not monitoring each cell voltage at an individual level. If you dump a cell in a 16sXp config, the BMS will not protect the assembly, it will keep chugging along if a cell fully dies.

Good point. Although, if I have each cell paralleled together on the balance cables, then if one totally died I would think it would then cause the voltage on that one cell to dip to 1/2 which would trigger a low voltage warning/action... right? [ie. if two cells are parallel, lets say one cell is 3.4v, the other dies and dips to 2v, the bms would see 2.7v which would trigger a fault if the delta between cells is too high since all the others would be at 3.4v]

MAybe you are right and I should just stick to one bms per 48v string...

 

[aentrop]

60A should be plenty; just make sure the BMS can parallel with the number you plan on using.

For JK BMS's you can add an unlimited number in parallel right? They don't talk to each other, they just monitor their own individual string of cells and if one of the bms's in parallel shuts down for some reason then it wont affect any of the others. Right?

Only problem I can see is if that bms tries to turn back on and its SOC is significantly different than the other strings then there would be a large transfer of current, depending of how far off the SOC is with the other strings. But if a BMS shuts off then an operator needs to come in and check the system anyways to see what error caused it to shut down.

 

[PreppenWolf]

Good point. Although, if I have each cell paralleled together on the balance cables, then if one totally died I would think it would then cause the voltage on that one cell to dip to 1/2 which would trigger a low voltage warning/action... right? [ie. if two cells are parallel, lets say one cell is 3.4v, the other dies and dips to 2v, the bms would see 2.7v which would trigger a fault if the delta between cells is too high since all the others would be at 3.4v]

MAybe you are right and I should just stick to one bms per 48v string...

If you put two batteries in parallel and one dies, nothing happens to the voltage measured across the pair. What will happen is the remaining battery will now be required to pass twice the current.

 

[sunshine_eggo]

I have 128 cells creating (8) 16s 48V packs. Each with a 100A BMS.

I personally would never mix S/P with individual cells. There is just too much risk of not monitoring each cell voltage at an individual level. If you dump a cell in a 16sXp config, the BMS will not protect the assembly, it will keep chugging along if a cell fully dies.

I do not disagree with your approach, just your stated outcomes and their associated risks.

The presence of a failed cell becomes very evident pretty quickly. In most cases, the failed cell will drain off capacity from it's parallel mate(s). This will create a notable drop in voltage that worsens over time. This results in a net loss of capacity that will manifest most often at charge when multiple OVP are triggered for light to moderatly cycled batteries. LVP trigger might be the first noticeable event.

If the battery is deep cycled, the above happens, but the problem will be noticed more quickly, and the first trigger might the the LVP.

When your 1P16S experiences a cell failure, it's exactly as the above two. It's just much more obvious.

The biggest risk is the nature of the failed cell. If it fails in a low resistance short, the failed cell may drain current at a very high rate with those associated risks.

 

[PreppenWolf]

On a 1p16s setup, the BMS is going to shutdown the pack immediately when a single cell exceeds parameters. My issue with the Xp16s setup is you are going to run a damaged pack longer making it more likely to damage other cells.

 

[sunshine_eggo]

On a 1p16s setup, the BMS is going to shutdown the pack immediately when a single cell exceeds parameters.

Same on 2P. It just may take longer.

My issue with the Xp16s setup is you are going to run a damaged pack longer

Not much longer.

making it more likely to damage other cells.

The only cells at risk is/are the one(s) in parallel with the failed cell, and that's in the case of the low IR short scenario above. The other series cell groups will be unaffected by the cell failure.

Again, not disagreeing with your configuration decision. Simply disagreeing with your risk assessment of parallel cells. IMHO, the advantages of single cell strings overwhelm the benefits of parallel cells - instant redundancy, knowing exactly which cell is having an issue, being able to take one or more of your batteries out of the system without taking the whole system down, multiple small BMS enabling a larger bank current, etc.

More BMS present also means more likelihood of BMS failure, and a failed BMS might take an entire pack out.

In my case, I have EV-grade Panasonic Li-NMC cells in a 21P14S configuration (and I have another identical battery to build). It would be astoundingly impractical for me to have 21 BMS (42 counting the second battery). Another risk mitigation factor for me is that NMC has strong voltage to SoC correlation. The flatness of the LFP voltage curve is going to mask any issues even with large SoC where my voltage deviations matter everywhere, and I'm going to notice it very very quickly.